ATLAS：为可靠世界模型规划对齐潜在结构传输

View PDF HTML (experimental)

Abstract:Latent world models rely on representation geometry for planning, yet regularizing the latent marginal alone does not determine the state-to-state relationships used for action selection. We show that this can cause planning-relevant novelty structure to be weakened as representations are transformed into the final latent used by the planner. We introduce Aligned Transport of Latent Structure (ATLAS), a training objective that explicitly preserves relational geometry while calibrating the global latent distribution. ATLAS transfers normalized pairwise structure from an informative encoder representation to the planning latent and uses Wasserstein embedding matching (WEMReg) to calibrate its marginal through one-dimensional Wasserstein-2 transport. Our analysis shows that relational preservation and marginal calibration impose non-redundant constraints, and connects finite-candidate planning stability to relational distortion, latent-scale mismatch, and prediction error. Instantiated in LeWM, ATLAS improves mean goal-reaching success across PushT, TwoRoom, and OGBench-Cube on both lower- and higher-novelty evaluation subsets, with the largest gain on higher-novelty TwoRoom episodes. Representation and rollout diagnostics further show stronger novelty-related structure in the planning latent, improved marginal calibration, and lower multi-step prediction error. Together, these results highlight preservation of planning-relevant latent geometry as an important ingredient for reliable world-model planning. Code is available at this https URL.

Subjects:	Robotics (cs.RO); Artificial Intelligence (cs.AI)
Cite as:	arXiv:2609.36333 [cs.RO]
	(or arXiv:2609.36333v2 [cs.RO] for this version)
	https://doi.org/10.48550/arXiv.2609.36333 arXiv-issued DOI via DataCite

Submission history

From: Yupu Yao [view email]
[v1] Mon, 28 Sep 2026 22:09:55 UTC (2,694 KB)
[v2] Wed, 30 Sep 2026 17:44:34 UTC (2,689 KB)